Regarding Reference Architecture & EA Repository

[aseth]

Hi,

I am seeking for some guidance on how can we create and manage Reference Architecture & EA Repository using Hopex. I would like to preserve Ref Architecture diagrams, patterns diagrams and decision tree in the repository.

Was wondering if you could help and guide here.

Thanks.

Amit Seth

[SergeThorn]

Hi Amit,

First when we refer to a Reference Architecture, we may associate them to the 4 (or 5) classical architecture domains or layers. It includes a document or set of documents that contains recommendations on the structures and integrations of IT products and services that make up a solution. It encompasses the best practices recognized in the industry, allowing it to suggest the optimal delivery method for this or that technology and in the design stages.

We will refer to generally four models

• Business Reference Model (BRM)
• Data Reference Model (DRM)
• Application Reference Model (ARM)
• Technical Reference Model (TRM) 

Enterprise reference architectures are standardized architectures that provide a reference framework for a domain (Business, Data, Application, Technology) or a vertical sector (ex: Banking-Finance, Insurance, Healthcare, etc).

Many domains have their reference architecture definitions.

Key reference architecture examples include:

• BIAN Services Landscape for the Banking Sector;
• ACORD Framework for the insurance sector;
• eTOM Business Process Framework for the Telecom Industry from TMForum;
• Government reference architectures, for example the American FEAF or the Australian AGA;
• Defense architecture frameworks such as NAF, DODAF and MoDAF;
• Reference architectures for manufacturing and supply chains such as ISA-95 and SCOR
• Etc..

I’m in the process of developing 2 reference models in Hopex:

• A BRM based on BIAN. For this one I’m describing the BIAN Service Domains into Business Capabilities and using Business Capability Maps. (Then I link applications, functionalities, etc…)
• A TRM using the Technical Services metaclasses and then producing a TRM Catalog.

Hope this helps.

Regards

[BenAvdicevic]

@SergeThorn , @aseth  this is very timely topic for our company too..

My 2 cents on above response.

BRM, TRM, DRM are "Reference Models" , not Reference Architectures. 

These two terms "Reference Architecture" and "Reference Model" are often used interchangibly - but they mean different things.

If you refer to TOGAF 9.2 (Chapter 41 - Repository)  you can see Open Group's caution on this.

Bottom line,

Reference Models are  (a set of unifying concepts about a specific domain/topic) 

- Think of a parts list when you buy IKEA furniture.

• Easy to do in HOPEX
  • BRM - Business Capability Map
  • DRM - Business Dictionary
  • TRM - Functionalities (probably technical service catalog is right here)

Reference Architecture are high level patterns ( solutions to a problem in context)

- Think of an IKEA diagram showing how parts fit together to make a chair

- Uses the concepts from the parts list (reference models) and adds relationships

• To me, its not immediately clear how to create generic reference architecture diagrams 
  • For example, imagine you want ot create Reference Architecture using a BRM (basically, a high level solution to a known problem - using only concepts defined in a capability map (e.g. BIAN)
  • What is your starting object to create such diagram containing, say 8 concepts from BRM?
• Do I create a new bus. capability MAP and call it "Reference Architecture ABC"
  • Then add only 8 concepts from BIAN map
  • Then create a Capability Structure Diagram for the new map?

For now, my prefererred approach would be to simply create reference architecture in Archimate - connecting Archimate Capability to the defined concepts in HOPEX capablity  map.    In Archimate, the viewpoint to chose is "Strategy".

Regards,

Ben

[hsoegaard]

I would say that any type of model you can make in HOPEX is potentially a Reference Architecture. Logical Application Architectures can be used both as a reference but also to model something that you are planning to implement. If you create a Resource and IT Architecture that is also a Reference Architecture. I would create my Reference Architecture in a specific Library and just use them for reference.. I dont think I would connect the reference architecture to capability maps etc. We are mostly using the capability maps to analyse impacts of the architecture and projects.

[aseth]

I agree to how @BenAvdicevic is thinking. Yeap they sound similar but they are different. Also @hsoegaard point is valid we should not associate a ref architecture with capabilities. But rather certain resources can implement a reference architecture or a pattern. I was able to realize this using the "Logic Application" meta-model and create a report out of it to make it visible and consumable to the end user interacting with the 360 portal.

[BenAvdicevic]

Thanks for feedbacks @aseth @hsoegaard .

Logical Applicaitons is something we've considred also.  However, its use is really only if you are trying to build a reference architecture with "Applicaiton or Application system" objects on your diagram.

With regards to capabilities.   This is a more business archtiecture concept.   IMHO, nothing wrong with creating a refernece architecture from business capabilities.

Yes - bus. capabilities are usually realized by a set of applications - so you could always translate a bus. capability diagram into one with applicaiton objects - in which case using HOPEX logical architecutre is a great fit. 

As for using a different library for reference architectures - it could work. We will be trying this out as well as using Archimate.

[aseth]

@BenAvdicevic The only thing I would like to bring up is the Diagraming under "Logical Application" and "Logical Application System" is not same as what you get under ""Applications" and ""Application System". It doesn't gives you the same realization since there isn't a real physical layer associated with that model. Associating capability is ok as long as its hidden on dashboards and reports where one would expect Applications to show up and not logical applications.

[aseth]

@ALL btw this is great conversation and learning. I wish we could get one a call on some monthly cadence. My next quest is to see how we could leverage HOPEX for EA Principles and have Application and Technology designs adhere to them. We could start a new thread too to get more visibility to this.

[hsoegaard]

Not sure which version you are using, but in HOPEX V5 Architecture Principles are now available, and can be used to connect to artifacts so you can see or check your compliance against the principles. Just in case you where not aware of it.

[BenAvdicevic]

@hsoegaard 

offtopic - but, which product is Architecture Principle associated with in HOPEX.

I can find the metaclass - but not much else